The effect of fuel formulation on the exhaust emissions of spark ignition engines

The Effect of Fuel Formulation

on the Exhaust Emissions of

Spark Ignition Engines

ARTHUR BELL

Dissertation presented for the Degree of Doctor of Engineering at

the University of Stellenbosch

Promoters : Dr A. B. TAYLOR & Prof. T.W. von BACKSTR M

April 2006

Declaration

I, the undersigned, hereby declare that the work contained in this dissertation is

my own original work and that I have not previously in its entirety or in part

submitted it at any university for a degree.

Synopsis

The research described in this dissertation examined the effects that fuel formulation can have on the regulated exhaust emissions produced by spark ignition engines in a South African context. Typical South African engine technology, and fuels representative of available fuels were investigated. To broaden the scope and provide information on as many fuel parameters as possible, fuel formulations other than typical retail fuels were also investigated. In order to gain insight into the mechanisms taking place, combustion analysis was performed on measured cylinder pressure traces. This type of analysis calculates the rate of combustion along with other useful parameters such as cylinder gas temperatures. A multivariate statistical analysis was then performed to enable the determination of the effects of the fuel formulation parameters of interest. This was done in such a way as to indicate the mechanisms through which the parameters influence the emissions.

An existing combustion analysis program was extensively modified as part of the research programme. The existing program consisted of a relatively simple single-zone combustion analysis while a two-zone combustion analysis model was added which splits the control volume into two distinct zones namely unburned reactants and burned products. An equilibrium reaction combustion model and routines for computing the gas properties of a mixture were incorporated. The extended Zeldovich NO formation model was also added to the combustion analysis routines to enable the investigation of some noteworthy statistical correlations identified in the research. The experimental results attained, as well as the results of the combustion analysis, were shown to be repeatable and significant. The combustion analysis was found to be a useful tool which was successfully used to explain the combustion related mechanisms that affected the measured emissions. The statistical approach used was sufficiently able to predict the fuel properties and combustion analysis parameters that influenced the emissions and the fuel properties that influenced the combustion parameters. In this way the mechanisms by which the fuel properties effect the emissions were explained. Many of the effects of the relevant fuel formulation parameters agreed with the observations reported in the literature considered.

The hydrocarbon emissions were seen to be mostly affected by factors which influence the post combustion burn-up and stoichiometry. Post combustion burn-up is either influenced by the amount of hydrocarbon containing mixture that is precluded from taking part in the bulk gas combustion process by storage and release mechanisms or by factors which influence the rate of the post combustion reactions. The response to the stoichiometry effects are well understood.

The oxides of nitrogen (NO) were found to be mostly influenced by fuel parameters which influence the combustion rates and the overall combustion timing: the location of 50% burned parameter was found to have good correlation with the NO emissions. The NO formation process relies on non-equilibrium, rate controlled reactions which are highly temperature dependent and fuel properties which cause the combustion to be advanced will result in higher temperatures and thus increased NO emissions.

Carbon monoxide (CO) emissions are seen to be influenced by stoichiometry and equivalence ratio effects and effects linked to the rate of change of pressure. The response to the stoichiometry and equivalence ratio effects are well understood, but the physical mechanism of the influence of the fuel parameters on the equivalence ratio

Opsomming

Die navorsing beskryf in hierdie verhandeling, ondersoek die moontlike gevolge van branstofformulasie op internasionaal gereguleerde uitlaatemissies van vonkontstekingenjins in ‘n Suid-Afrikaanse konteks. Tipiese Suid-Afrikaanse enjintegnologie en brandstowwe verteenwoordigend van brandstowwe beskikbaar in die handel, is ondersoek. Brandstofformulasies verskillend van tipiese beskikbare brandstowwe is ook ondersoek om die omvang van die projek te vergroot en soveel moontlik inligting ten opsigte van brandstofparameters beskikbaar te stel. Verbrandinganalise is op gemete verbrandingsdrukdata uitgevoer om die verbrandingsmeganismes wat voorkom, beter te verstaan. Die tempo van verbranding en verbrandingsruimgastemperatuur is van die uitsette wat bereken word met behulp van verbrandingsanalise. ‘n Multiveranderlike statistiese analise om die effek van branstofformulasieparameters te bepaal, is op so ‘n manier uitgevoer dat die meganismes waardeur hierdie parameters emissies beinvloed, aangetoon word.

Die oorspronklike program, ‘n relatief eenvoudige enkelsone-verbrandingsmodel, is omvattend gewysig om ‘n tweesone-verbrandingsanalisemodel daar te stel wat die verbrandingsvolume in twee sones - onverbrande reaktiewe en verbrande produkte -verdeel.’n Ekwilibriumreaksie-verbrandingsmodel en subroetines om die gaseienskappe van ‘n mengsel te bepaal, is ook geinkorporeer. Vir die bestudering van sekere interessante statistiese korrelasies wat gedurende die navorsing opgemerk is, is die uitgebreide Zeldovich stikstofmonoksied (NO) formasiemodel ook by die analiseroetine gevoeg.

Die eksperimentele resultate asook die resultate van die verbrandingsmodel is as herhaalbaar en beduidend aangetoon. Die verbrandingsanalise is ‘n bruikbare werktuig wat suksesvol aangewend is om te verduidelik hoe die meganismes wat verband hou met verbranding die gemete uitlaatgasemissies beinvloed. Die statistiese metodes kon aanvaarbare indikasies gee watter brandstofeienskappe en verbrandingsanalise-parameters emissies beinvloed en ook watter brandstofeienskappe verbrandingsparameters beinvloed. Op hierdie manier is die meganismes waardeur die brandstofeienskappe ‘n uitwerking op uitlaatgasemissies het, verduidelik. Die uitwerking van die relevante brandstofformulasieparameters stem grootliks ooreen met die waarnemings aangemeld in die bestudeerde literatuur.

Die analise het aangedui dat onverbrande koolwaterstofemissies die meeste beinvloed word deur faktore wat ‘n uitwerking het op stoichiometrie en na-verbrandingsoksidasie. Na-verbrandingsoksidasie word deur die volgende beinvloed: die hoeveelheid mengsel wat koolwaterstofverbindings bevat wat nie deelneem aan die hoofverbrandingsproses as gevolg van opgarings- en verspreidingsmeganismes nie of deur faktore wat ‘n uitwerking het op die tempo van na-verbrandingsreaksies. Die reaksie tot stoichiometriese effek is duidelik. Daar is gevind dat stikstofmonoksied (NO) primer beinvloed word deur brandstofparameters wat op verbrandingsgastemperature inwerk. ‘n Toename in globale verbrandingsgastemperature lei tot ‘n vermindering in gevormde NO wat moontlik ‘n gevolg is van die toenemende NO ontbinding wat plaasvind omdat NO verbindings- en ontbindingsreaksies ‘n nie-ewewigtige, tempo-gereguleerde patroon volg.

Daar is gevind dat die stikstofmonokside (NO) hoofsaaklik beïnvloed word deur brandstofparameters wat ontbrandingstempo”s en algehele ontbrandingstydreëling beïnvloed: ‘n goeie korrelasie is gevind tussen die posisie van die 50% brandpunt en die NO uitlating. Die NO formasieproses is afhanklik van nie-ewewigtige,

tempogereguleerde reaksies wat hoogs temperaturafhanklik is. Brandstofeienskappe wat veroorsaak dat die ontbranding vroeër plaasvind, sal hoër temperature en dus verhoogde NO formasie tot gevolg hê.

Daar word gesien dat koolstofmonoksied (CO) emissies deur stoichiometrie en ekwivalensverhouding beinvloed word; asook deur oorsake gekoppel aan die tempo van verandering in verbrandingsdruk. Die reaksie op stiochiometrie en ekwivalensverhouding is duidelik, maar die fisiese meganismes van die invloed van brandstofparameters op ekwivalensverhouding is onduidelik.

Dedication

I dedicate this to my parents for all that they have done for me, to Janet, my

ever patient wife, for all the love and support that I could wish for and to Jarryd,

our son, who brightens every day.

Acknowledgements

Dr Andrew Taylor, for all the encouragement and guidance throughout this project.

Professor von Backström for always encouraging me to finish as well as his input in the final

stages.

Sasol Oil, Research and Development, for financial and technical support including the

supply and laboratory analysis of the test fuels.

The Technology and Human Resources for Industry Programme (THRIP) of the National

Research Foundation (NRF) for financial support.

Toyota South Africa, for the donation of the engines used in the research.

Derek Moran for all the assistance with the programming, for making this seem like fun at

times, and for his valued friendship.

Dr Randall, for the assistance with the statistical analysis.

Benn Vincent and the rest of the staff at the Centre for Automotive Engineering, for the

assistance with the engine testing.

Table of Contents

! ! " # ! $ ! % &$ ' ( & $ %

) * + , ' ( $ ( - &. " " - & & $ /

) % - / ) ) 0 1 23 3 2 3 . 3 " 4 ) ) % 23 3 5 ) ) ) 6 3 0 & %7 ) / " !3 , %% ) / % " - 3 %) ) 8 " !3 " - %9 ) 8 % " %9 ) 8 ) 3 %5 ) 8 / * 3 %: ) 8 8 - )7 / & " ' ( ' " ' " . &" ; )% / % 3 " ' " )% / % % 3 0 &' 2 )% / % ) 3 " 2" " )4 / ) 3 & 3 " 2 " ): / ) % & " !3 " ): / ) ) . " " !3 " /7 / ) / " " !3 " /7 / / /9 / / % " 6 " /9 / / ) ' 2 . 6 " /5 / / / ' 6 " /5 8 ' " " ' 6 * < " &. " $ - " 6 , = 87 4 ' " " ' 6 ( " 88 4 % 88 4 ) . 89 4 / $ 85 4 8 > 6 > ! > $ ' 85 4 8 % > 6 ' 8? 4 8 ) ! - 8: 4 8 / $ - 8: 9 " @ 2" ' &- " $ 22 ' ( 47 9 % " " ! A 47 9 % % 8 A " " A 47 9 % ) 8= " A 4) 9 ) 2 - 4) 9 / " A , 44

9 8 ! " # 45 5 " @ 2" ' &- " $ 2' " ( ' " 4? 5 % # 4? 5 ) " 2 4? 5 / 2 & 97 5 8 2 97 5 4 9% ? 6 " ' " " " $ = & 9/ ? % 6 ' * ' 9/ ? % % 2 ' 6 ' 9/ ? % ) 6 ' 6 ' 9/ ? % / * & 2 98 ? ) AB 6 ' - 94 ? / AB 6 ' - 94 ? / % 3 AB 3 - 99 ? / ) , 2 9? ? / / " # 2 - 9: ? / 8 6 A- 58 ? / 4 $ - 55 ? / 9 3 ; ?7 : - * ( & $ $ = & ?/ : % 2 ' ?/ : ) 2 & ?9 : ) % & ?9 : ) ) AB :9 : ) / AB ::

%7 & & $ = & %7 )

%7 % - ! %7 ) %7 ) 3 %7 / %7 / 3 - %7 8 %% ' " ( %7 ? %% % ' %7 ? %% % % ' 2 " %7 ? %% % ) ' " ' %7 : %% % / ' ' %%) %% ) 2 " ' %%/ %% / . AB - %%8 %% 8 . B 3 $ %%? %% 4 ' %)% %% 4 % 6 " ' %)% %% 4 ) $ " ' %): %% 4 / " ' %// %% 4 8 ' C 2 %/5 %% 9 $ %89 %) & ( & $ ' " ( %44 %) % 6 " %44 %) % % %44 %) % ) ! %45 %) % / . %4?

%) ) $ " %9% %) ) % %9% %) ) ) . %9) %) ) / %9/ %) / " %9/ %) / % . %9/ %) / ) ! %98 %) / / %98 %) 8 " " %94 %/ $ ( & $ %99 %8 ' " - - " $ & $ %57 %4 ' " " ' " $ " %5% 22" $ &@ > $ = & ' " " ' 6 ( " %

22" $ &@ * > " ' &. & $ ; B $ " 6 " ' " " " ' - ( , &. " $ * = + ' &" , " ' $

* ' - $ * %

22" $ &@ > " ' &. & $ 6 " ' $ " ' " & &" $ " 0 ( & $ %

22" $ &@ > * ( " ' " ( %

22" $ &@ " >, ' 26 & = 2' " " $ " " - & & $ ' " ( " % 22" $ &@ > ' " ( & & ' " , ' " & $ - " &$ , %

List of Figures

)A% " $ ! $ - 6 D E5F 9 )A) G H ' 3 3 G H $ ! 3 ! 1 3 E4F 5 )A/ 3 6 1 3 < %::9 G 3 - H : )A8 3 , 1 %::9 G 3 - H %7 )A4 - # 3 # %) )A9 G H - 2 1 ! G H 3 G H $ 3 G E%9FH %4 )A5 ! 3 # # 3 E%?F %? )A? 2 54 3 3 3 3 3 E%?F %?

)A: " " ' %4A7 8 E%?F %:

/A% D 3 " 2" " E)7 F )9 8A% & 3 3 3 8% 4A% 3 ! 3 3 3 3 3 3 84 4A) 3 ! 3 3 3 3 3 84 4A/ . 3 3 85 4A8 6 > ' $ G , E/)FH 8? 9A% 23 3 3 3 3 $ % # 3 ! 3 3 3 48 9A) 23 3 3 3 3 3 3 G 3 3 3 H 3 A 3 3 1 3 3 48 9A/ 23 3 3 3 3 3 $ )C $ 1 3 1 $ 1 3 ) 1 ) 6 ) 3 3 3 3 3 3 49 ?A% 3 ' 3 8 A " 98 ?A) G4 :4I ! H 94 ?A/ 3 3 AD 1 99 ?A8 3 ' ) ?) :A% " ! 3 3 3 3 3 ?8 :A) ' 2 ! ?5 :A/ ! 3 G/7 7 7

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

%%A)/ 3 3 ! %8: %%A)8 3 3 ! %4% %%A)4 %4) %%A)9 $ ! G - @ H %4) %%A)5 3 47 I 3 %4/ %%A)? 3 ! G - @ H 3 47 I %48 %%A): 3 3 $ 3 47 I %48

!

List of Tables

/A% 2 & 2 3 0 &' 2

)) /A) 2 3 ( 3 3 3 0 &' 2 )/ /A/ 3 1 - * 7 : " # ' + E)8F /% /A8 3 - * + E)4F /8 8A% 3 8/ 4A% - 1 ! 3 84 4A) - ! 1 D 1 3 1 1 ! 1 3 3 89 4A/ - 1 ! 3 3 85 4A8 - 1 ! 3 1 ' $ - $ 85 9A% " 4) 5A% " 1 # 1 & 2 " # ' 3 2 3 8 A " 4: 5A) " # 3 2 3 8= 97 ?A% ' 3 ! B 3 3 3 5: :A% 3 ! 3 3 ?4 :A) ! ; 3 E4/F :) :A/ 1 1 & 3 3 ; 3 3 :9 %7 A% 3 %7 8 %7 A) M 6 %7 4 %7 A/ M 3 6 " 1 3 %7 9 %7 A8 ! 3 A 3 %7 5 %%A% 3 %%7 %%A) 3 3 3 3 %%7 %%A/ 3 3 AD %%) %%A8 3 3 AD %%/ %%A4 3 3 3 3 3 AD %%/ %%A9 3 3 3 3 3 AD %%/ %%A5 6 L C 2 %)) %%A? A 1 3 A 3 3 6 L C 2 %)/ %%A: 3 3 6 L %)4 %%A%7 3 6 L 3 3 %)9 %%A%% & A A 1 3 1 6 L %)5 %%A%) 6 L C %)?

%%A%/ A 1 3 A 3 3 6 L C %): %%A%8 $ C 2 %/7 %%A%4 A 1 3 A 3 3 $ C 2 %/7 %%A%9 $ C %/% %%A%5 A 1 3 A 3 3 $ C %/) %%A%? $ 1 8 %/) %%A%: C 2 %// %%A)7 A 1 3 A 3 3 C 2 %/8 %%A)% & A 1 3 1 %/4 %%A)) C %/9 %%A)/ A 1 3 A 3 3 C %/5 %%A)8 2- @ C 2 %/? %%A)4 A 1 3 A 3 3 2- @ C 2 %/? %%A)9 - @ C 2 %/: %%A)5 A 1 3 A 3 3 - @ C 2 %/: %%A)? & A 1 3 1 - @ %87 %%A): 6 ' - @ C 2 %87 %%A/7 A 1 3 A 3 3 6 ' - @ C 2 %8% %%A/% & A 1 3 1 6 ' - @ %8% %%A/) * 4:4 C 2 %8% %%A// A 1 3 A 3 3 * 4:4 C 2 %8) %%A/8 & A 1 3 1 * 4:4 %8) %%A/4 B )* ):? C 2 %8) %%A/9 A 1 3 A 3 3 B )* ):? C 2 %8/ %%A/5 & A 1 3 1 B )* ):? %88 %%A/? & 3 3 A 1 3 1 B )* ):? %88 %%A/: " # ' C 2 %88 %%A87 A 1 3 A 3 3 " # ' C 2 %84 %%A8% & A 1 3 1 " # ' %89 %%A8) & 3 3 A 1 3 1 " # ' %89

! %%A88 3 %%A)/ %47 %)A% 3 6 3 0 &' 2 " 2" " 3 3 %49 %/A% 6 %95 %/A) $ %9? %/A/ %9:

List of Equations

" # 4A% ! ! 8: " # ?A% " # 3 3 AD G ! * H 95 " # ?A) 95 " # ?A/ 3 3 , - * E/%F 3 9? " # ?A8 - ! 9: " # ?A4 " # 3 3 3 9: " # ?A9 3 # 3 3 # 57 " # ?A5 3 3 # 3 # 57 " # ?A? 3 # 5% " # ?A: - 5% " # ?A%7 3 3 # 5) " # ?A%% ! 3 A # 5/ " # ?A%) 3 A # 3 5/ " # ?A%/ 3 3 ; 3 54 " # ?A%8 3 3 54 " # ?A%4 ' 3 ! B 3 3 3 $ 5? " # ?A%9 ' $ 5? " # ?A%5 ' 5?

" # ?A%? ' " # ?A%9 3 " # ?A%5 5?

" # ?A%: " # 3 # # 1 3 # $ 5: " # :A% " # ?? " # :A) 3 # ?: " # :A/ γ 3 6 E8)F ? 4I ?: " # :A8 " # 3 :4 " # %7 A% 3 ! # 3 %7 /

!

Nomenclature and Abbreviations

° CA... Degrees of Crank Angle Rotation

ACEA ... Association des Constructeurs d’Automobiles (European association of automobile manufacturers

Alkane... Paraffin – an alternative name used by some authors Alkene... Olefin – an alternative name used by some authors

Burn Angle ... Number of degrees of crank shaft rotation required for combustion to proceed from a specified start point, to a specified end point.

Burn Rate... Rate of combustion in internal combustion engines

Catalyst Light Off ... Time taken for the exhaust catalyst to warm sufficiently that efficient operation is achieved

CI ... Compression Ignition (refers usually to diesel engines) CO... Carbon Monoxide

CO2... Carbon Dioxide

CVS... Constant Volume Sampler DOHC ... Double Overhead Camshaft

E##... Distillation Point: % of fuel vaporised at ## °C, [%]

EMS ... Engine Management System

Engine-out emissions ... Pre-catalyst exhaust concentrations (if fitted, or else expected tail-pipe emissions if no catalyst fitted)

EPA... Environmental Protection Agency (North American regulatory body) Equivalence Ratio ... Operating fuel/air ratio divided by stoichiometric fuel/air ratio

ER (φ)... Equivalence Ratio Calculated as [operational fuel/air ratio]/[stoichiometric

fuel/air ratio] ( ER > 1 - rich, ER < 1 - lean)

ETA... Engine Test Automation – PC based data acquisition system developed in the Department of Mechanical Engineering, University of Stellenbosch

EUROPIA... European Petroleum Industry Association

EVO and EVC... Crank angle at the point of Exhaust valve opening and closure respectively Flame Quench Distance ... Minimum distance for a given geometry, through which a flame will propagate FTP ... Federal Test Procedure

Fugacity ... Gas property used in real gas thermodynamic calculations HC... Various Hydrocarbon Species

HFID... Heated Flame Ionisation Detector, gas analyser type for measuring hydrocarbons

IBP ... Initial Boiling Point, [°C]

IVO and IVC... Crank angle at the point of Inlet valve opening and closure respectively Lambda (λ) ... Inverse of Equivalence Ratio (λ < 1 - rich, λ > 1 - lean)

Lambda Sensor ... Exhaust Oxygen Sensor, intended to measure operational air/fuel ratio, or Lambda

Modal Analysis... Measurement of instantaneous exhaust emissions during driving cycle test - allows the identification of which components or “modes” of the driving cycle correspond to peaks in emissions

MTBE... Methyl Tertiary Butyl Ether

NDIR ... Non-Dispersive Infra Red, gas analyser type (can measure many gas species) NMHC ... Non-Methane Hydrocarbons, Various Hydrocarbon Species, excluding

Methane (Methane occurs naturally and is relatively inactive in photochemical reactions, and is therefore often ignored)

NOx... Oxides of Nitrogen

OE... Original Equipment

OEM... Original Equipment Manufacturer

Photochemical Reactions ... Chemical reactions catalysed by ultra violet radiation

Racer ... Rapid Acquisition of Combustion Engine Results - PC based high-speed cylinder pressure data acquisition and combustion analysis system developed in the Department of Mechanical Engineering, University of Stellenbosch Regulated Emissions ... CO, HC and NOx, exhaust gases covered by legislation in developed countries SI... Spark Ignition (refers usually to petrol engines)

SOHC... Single Overhead Camshaft

Stoichiometric ... Ideal fuel/air ratio, exactly sufficient fuel supplied to an engine that it combusts

all the induced air to form only CO2 and water

T##... Distillation Point: Temperature required to evaporate ##% of fuel, [°C]

Tail-pipe emissions ... Post-catalyst exhaust emissions TDC ... Top Dead Centre

Vapour/Liquid Ratio (V/L Ratio) ... Measure of volatility, directly related to vapour lock problems VOC ... Volatile Organic compound

%

1. INTRODUCTION

The importance of gaseous pollutants associated with automotive emissions has long been known and much research has been undertaken in an attempt to reduce the impact of these on air quality. Most developed countries have strict legislation limiting the emissions of certain gaseous compounds of new vehicles, and these limits are continuously being lowered as technology improvements allow engineers to produce lower emitting vehicles. While South Africa does not at present have any passenger vehicle gaseous emissions legislation, Cape Town and other cities are known to, at times, have poor air quality. The role of automotive related emissions in this poor air quality is undoubtedly important and the first steps in the implementation of vehicle emissions regulation have recently been taken. Legislation equivalent to ECER83.04 which is commonly referred to as Euro 2, comes into force for new vehicle homologations from 1 January 2006, and will be applied to all newly manufactured vehicles from 1 January 2008 [1]. It is intended that the legislation will gradually be tightened to bring it in line with the later ECE regulations. Vehicle emissions regulation is the most important mechanism to reduce vehicle emissions, and thus air pollution, however the effect of the fuel formulation may also be important. Furthermore, during the transition phase when most of the vehicles on the road were manufactured prior to the legislation enforcement dates, and are thus uncontrolled, any potential reductions in emission from these vehicles may have a significant impact on air quality. Thus it is prudent to investigate the potential for influencing emissions through alterations to the fuel formulation as these effects can be felt immediately. The objective was thus to investigate whether meaningful differences in exhaust emissions can be achieved through fuel reformulation.

Vehicles impact air quality predominantly through two sources of pollutant emission. The first source is exhaust emissions, which are gaseous or particulate emissions released as a result of the combustion of fuel by the engine in the vehicle. The second source is evaporative emissions, which is the gaseous release of hydrocarbon compounds evaporated from the fuel storage and supply system. This research is concerned with the first source of emission, the exhaust or tail pipe emissions. Hydrocarbon emissions, both exhaust and evaporative, and the oxides of nitrogen emitted in the exhaust are precursors to photochemical smog and ozone formation. This is one of the major sources of poor air quality associated with vehicles. As the hydrocarbons are emitted in both exhaust and evaporative emissions, they are obviously of great importance. It must therefore be stressed that any action taken to attempt to reduce the air quality impact of the passenger vehicle must consider both sources of vehicle related emissions. It is probable that the main source of hydrocarbons emitted by non-regulated vehicles such as those making up the majority of those on the South African roads is evaporative. At the time of inception of this research, the introduction of unleaded fuel to the South African market was imminent and there was much speculation as to the potential impact of this on air quality. This had some bearing on the choice of the fuels, and their specific formulation, used in the research. The objectives of the research were to study the effect of the formulation of the fuel on spark ignition engine exhaust emissions. The entire scope of the related work included the study of the regulated emissions (hydrocarbons, oxides of nitrogen and carbon monoxide) and speciated aldehydes from a range of fuels. The fuels included many of the market fuels being produced at the time, as well as unleaded formulations that were proposed for introduction. A number of

fuels were also specifically formulated to investigate the effects of a variety of fuel parameters. In all, thirty-five fuels were included in this phase of the research. A related study also investigated the effect of a number of alcohol compounds as potential fuel blending components in varying concentrations. This dissertation is limited to the regulated emissions of the thirty-five leaded, un-leaded and specially formulated fuels. The objectives of the study were to attempt to identify the fuel formulation parameters that influence the emissions, and to gain some understanding of the mechanisms through which the effect is occurring.

Vehicle technology and engine design influence exhaust emissions to a great extent. In order to meet the strict vehicle emissions legislation it is necessary to utilise exhaust gas after treatment. Homogenous charge, spark ignition engines are the dominant technology in passenger car and light commercial vehicles and this technology requires the use of a three-way catalytic converter. The use of this device also places constraints on the engine technology and operating parameters. The majority of vehicles on the South African roads, and of new vehicles currently being sold, do not have this technology and therefore are not constrained in their engine design. For this reason some local engines utilise outdated technologies or a mix of current technology without being constrained to use engine calibrations necessary for exhaust catalytic converter operation. This is one of the main reasons why this research was necessary when so much research had been performed internationally, as the results achieved in these other programmes are often based on the newer technology engines. For this reason engines typical of the technologies predominantly in use in the country were chosen.

The approach taken in terms of the fuel formulation and fuel formulation parameters investigated was somewhat different to the typical approach. In most previous studies a small number of fuel formulation parameters are pre-selected and a full factorial experimental approach is followed. In this research a large number of fuels, spanning very many different levels of a large number of fuel formulation parameters of interest, were blended and tested. The highly complex and interdependent nature of the process of exhaust emissions formation implies that the determination and quantification of the important parameters would be difficult. For this reason, statistical methods utilising multivariate linear regression techniques were used. The statistical analysis was constructed so as to enable the identification of the mechanisms through which the fuel parameter influences the emissions produced.

One of the main objectives was to gain an understanding of the mechanisms through which the fuel properties were influencing the emissions produced. This would require knowledge of the physical processes taking place and the interaction of the fuel properties with these processes. Advanced combustion analysis was identified as being necessary to achieve this. Combustion analysis is the process whereby the rate of combustion is determined by measuring the cylinder pressure trace in an engine, and performing specialised thermodynamic analysis thereon. This information can then be used to compare the combustion characteristics of the different fuel formulations. Although an existing software program was available for this purpose, the thermodynamic analysis was thought to be too simplistic and therefore a more advanced analysis was required. This was added to the software and used, along with the original analysis, in the analysis of the results. A further major modification was made to the thermodynamic analysis in order to investigate the cause of a noteworthy finding from the initial statistical analysis showing an unexpected interaction between the fuel properties and the exhaust NO emissions. This analysis was able to provide valuable insight into the mechanism of NO formation and thus

/

2. BACKGROUND TO AUTOMOTIVE EMISSIONS

2.1.

Motivation

The important influence of automobile exhaust emissions on air quality, and in particular on photochemical smog, has been known for many years. Two of the major precursors in the formation of photochemical smog are present in significant concentrations in the exhaust gas of Spark Ignition (SI) engines, namely the Oxides of

Nitrogen (NOx) and Hydrocarbons (HC). Photochemical smog is of great concern in some highly populated cities

around the world, the best known example being Los Angeles (LA). Automobile exhaust emissions were discovered to be a major source of the pollutants causing this and for this reason legislation was introduced in the United States of America (USA) in the 1970’s which limited the exhaust emissions of new vehicles. Since then, the USA and many other developed countries have introduced increasingly stricter legislation in an attempt to improve air quality. In South Africa, Cape Town too is thought to have a photochemical smog problem, which is exacerbated by a peculiar geographical and meteorological phenomenon, which results in temperature inversion. This results in warm ground air being trapped below a layer of cooler air which prevents dispersion of the pollutants, intensifying the formation of photochemical smog. Temperature inversion occurs in Cape Town mostly on wind-still autumn and winter days resulting in visible smog, often called brown haze. A limited study incorporating a source apportionment of the Cape Town Brown Haze was undertaken and the results indicate that the major contributor of the visibility degradation is diesel engine particulate emission [2]. However, studying detailed ambient pollutant concentration data monitored at various locations in Cape Town by the Air Pollution Control and Scientific Services divisions of the Cape Metropolitan Council (CMC) indicates significant photochemical activity associated with gaseous vehicle emissions [3]. Air quality and photochemical smog are discussed in more detail below in Section 2.2.

The strict legislation introduced internationally has forced vehicle manufacturers to seek means of reducing the tailpipe emissions from their automobiles. This has resulted in much research being done internationally in an attempt to learn more about the mechanisms of automobile emissions formation and the factors that influence them. The majority of this research has been done in the USA where legislation is of the strictest in the world. The result of much of this research has led to major modifications to the SI engine itself in the last few decades, the most significant being electronic management of fuel injection and ignition timing. This allowed improved control of air/fuel ratio which can have a substantial effect on the exhaust emissions produced by the engine. However, development of the engine itself for improved emissions could not meet the strict legislation. This then necessitated the addition of exhaust after-treatment methods, the most significant being catalytic converters, which reduce the pollutants to legislated levels. Advanced three way catalytic converters only operate efficiently over a narrow equivalence ratio band, which necessitated the development of improved closed-loop air/fuel ratio control. This improved control mechanism employs an oxygen sensor, often called a lambda sensor, placed in the exhaust which provides feedback to the electronic engine management system (EMS). As catalysts and lambda sensors are poisoned by the lead based additive used as an octane improver in conventional gasoline,

new unleaded gasoline blends had to be developed. Further recent developments have included the production of direct injection, stratified charge, engines usually referred to as Gasoline Direct Injection engines (GDI). A significant body of research exists aimed at quantifying the effects of the fuels formulation on the exhaust composition. This has led to the implementation of legislation in the USA which requires Reformulated Gasoline (RFG) to be available in certain regions where certain air quality standards are not met. Furthermore, as legislation is expected to get ever stricter, improvements to conventional engine technology and after treatment need to be supplemented with improved fuels to meet these targets. This has also promoted international research into the effect of fuel formulation on gaseous emissions.

In South Africa, legislation which will enforce the use of catalytic converters will only come into force, in a phased approach, starting in January 2006 [1]. Therefore local vehicle manufacturers have, up until now, been able to utilise traditionally lower cost engine technologies that have become outdated internationally. Unleaded gasoline became available in March 1996, and leaded gasoline will only be removed from fuel supply in January 2006. This lack of availability of unleaded fuel prior to 1996 had forced the utilisation of predominantly open-loop control of air/fuel ratio, as the sensor required for closed loop control, the lambda sensor, is poisoned by the lead. It is also important to note that the South African vehicle fleet differs in another significant way from fleets in other developed countries in that, due to the lack of emissions legislation, there are no evaporative emission controls on many of the models. This, combined with two significant factors, make the level of hydrocarbon evaporative emissions very large from these vehicles. Firstly, the utilisation of modern fuel injection systems, on some vehicles, results in the fuel being pressurised to high pressures (up to and over 300 kPa) for injection accompanied by high levels of fuel circulation (100 litres/hr). This results in the fuel being heated to a larger extent than with older technology carburettor equipped vehicles, leading to increased evaporation. Secondly, the fuel specifications internationally have tended towards fuels with reduced vapour pressure to minimise the evaporative potential of the fuel. The specifications in South Africa have not followed this trend and appear to be similar to specifications in countries with much cooler climates. All of these factors result in the evaporative emissions of modern South African vehicles being much greater than that of vehicles fitted with the evaporative control devices and this has been shown by de Waal [4] and van der Westhuizen [5]. The raw fuel lost to evaporation will also take part in the photochemical reactions and may conceivably be more significant than the exhaust hydrocarbons.

South Africa is also in a unique situation as regards the range of gasoline formulations available. With the large difference in altitude between the coastal and inland regions, four different fuel octane grades are available, both regions having a high and low octane grade. Fuel is also produced locally in a number of different ways producing unique fuel formulations. Fuel from coal and natural gas is produced by Sasol, while PetroSA (formerly Mosgas) produces fuel only from natural gas, both companies utilising variations of the Fischer-Tropsch process. Crude based fuel however, remains predominant, being produced by a number of refineries around the country.

A further motivation for studying the effects of fuel formulation on emissions is that South Africa has a low turnover of new vehicles by international standards. This means that the average age of the vehicles on the road

4

introduced will take considerably longer to filter down into the vehicle park. However, if any improvement in emissions can be achieved through an alteration in fuel formulation, an immediate benefit is possible.

All of the above factors imply that results achieved internationally may not be directly translatable to the local conditions. At the time of inception of this research project all the locally produced fuel contained lead-based additives, while the production of unleaded fuel was imminent. It was therefore decided that a study of the emissions implications of the local leaded fuel formulations and that of the possible unleaded formulations, operating in engines representative of those currently being manufactured locally, be undertaken.

An important aspect of the research was the intention to study and quantify the mechanisms through which the fuel formulation influenced the engine out emissions. This fundamental understanding of these mechanisms would provide a direct transference to new vehicle technologies as well as being relevant to fuels manufactured under different regulations (such as after lead phase out). Furthermore, the experimental study incorporated aspects intentionally designed to simulate the vehicle technologies associated with exhaust after treatment engine technologies.

2.2.

Air Quality, Photochemical Smog and Pollution Associated with

Vehicle Emissions

The major impact that automobile exhaust emissions have on air quality and pollution is in the form of photochemical smog. Smog is a term originating in England (about 1911) as a synonym for the mixture of fog and coal smoke. Smog is thought to arise from the formation of sulphur trioxide, which forms hygroscopic nuclei, which, after absorbing water, forms sulphuric acid. Smog occurs on cold, wet, winter days or nights with low ozone concentrations and low visibility. Photochemical smog, on the other hand, occurs on hot, dry, summer days with high ground level ozone concentrations, and reduces visibility to a lesser extent. The process of photochemical smog formation is known to rely on photochemical reactions. The most popular theory [6] is that particular mixtures of hydrocarbons and nitrogen oxides react in the presence of, and are catalysed by, ultraviolet radiation to form a variety of products including aldehydes and ozone, which is a powerful irritant. Caplan [7] theorises that the formation of photochemical smog is governed not only by the quantity of hydrocarbons but by their reactivity, which he defined in terms of the affinity to form smog when present with nitric oxide. His data has indicated that smog is always reduced by decreasing the concentrations of reactive hydrocarbons, and that for a given concentration of hydrocarbons, the formation of smog is at a maximum for one particular nitric oxide concentration. This implies that the reduction of nitric oxides in the atmosphere may conceivably increase smog

formation. This is supported by Amann [8] who defines three separate scenarios: the NOx inhibition region, the

knee region and the HC saturation region as shown below in Figure 2-1. In the HC saturation region reduction in

HC concentration has no effect on ozone formation. In the knee region decreases in either NOx or HC will

reduce the ozone formed. However, in the NOx inhibition region decreasing the HC will reduce the ozone formed,

but independently reducing NOx may actually increase the formation of ozone. Thus the effect on ozone

formation from reduced vehicle emissions is dependent on the initial composition of the local urban atmosphere. This peculiarity is due to the fact that in the group of chemical reactions that Caplan postulates to govern smog

formation in its role as a precursor to ozone formation and in reaction 11 it inhibits smog formation by reacting with oxyalkyl radicals and thus terminating the chain reaction.

Understanding of this interplay between the relative concentrations of NOx and HC is very important. Using the

situation in Cape Town as an example, it is possible that prior to the introduction of unleaded petrol, many vehicles had high compression ratio engines, but no exhaust after treatment. Such cars are likely to be relatively

high emitters of NOx. The introduction of unleaded petrol then enabled the introduction of vehicles with catalytic

converters, and many up-market vehicles, fitted with these devices, have been sold locally even in the absence of legislation. These vehicles will be low overall emitters. The mix of vehicle technologies on the road has thus changed in the last ten years. It is therefore conceivable that previously Cape Town may have been sitting in the

NOx inhibition region (upper left of Figure 2-1) due to a relatively high number of high NOx emitters being on the

road, and that any reduction in emissions would move the relative mix of NOx and HC. Reducing emissions from

the NOx inhibition region, either vertically down due to NOx reduction only, or diagonally due to overall reductions,

will have the tendency to initially worsen the situation as it enters the knee region. Improvements would only be seen once the HC saturation region is reached.

An indication of the differences in the reactivity of different hydrocarbon groups can be seen in Figure 2-2 (b)

where it can be seen that substituted internal olefins are extremely reactive whereas the paraffin family and benzene ring compounds are much less reactive.

5

Figure 2-2 (a) Routes for photochemical smog formation. (b) Nitric oxide photo oxidation rate, or reactivity of different hydrocarbon classes [6].

Other undesirable compounds that may be linked to automobile exhaust emissions, and classified as pollutants,

would include CO2, CO, aldehydes, certain potential carcinogenic hydrocarbons, sulphur dioxide, particulates as

well as compounds containing lead. CO2 is classified as a “greenhouse gas” as CO2 in the atmosphere is

thought to block the radiation of heat away from the planet and thus, by trapping this heat, is causing a climatic change by increasing global average temperatures. The haemoglobin in the blood has a higher affinity for CO than oxygen, thus any CO inhaled bonds with the haemoglobin and is only slowly replaced by oxygen in the lungs. This reduces the ability of the body to exchange gases and causes drowsiness, impairs alertness, thinking and reflexes and can, in the extreme, be fatal. There are other compounds found in the exhaust that may be of interest due to their harmful nature. Aldehydes are irritants and some are thought to be carcinogenic while Benzene is a known carcinogen. Another carcinogenic compound that has received some attention lately

is 1,3 Butadiene. Sulphur dioxide (SO2) oxidises to form sulphur trioxide (SO3) which combines with water

vapour to form sulphuric acid (H2SO4). Sulphur dioxide is a strong irritant and sulphuric acid forms acid rain.

Particulate matter can potentially adsorb carcinogenic hydrocarbons and carry these deep into the lungs where they can cause serious damage. Particulate matter is also known to reduce visibility and is a contributor to visible haze. The octane enhancing additive tetraethyl lead, used in gasoline since the 1920’s, produces emissions of various compounds containing lead: these compounds are known neurotoxins and are thus undesirable.

2.2.1.

Photochemical Smog in Cape Town

During episodes of temperature inversion, a meteorological phenomenon in which a layer of warm air is trapped below a layer of cooler air, there is a tendency for a haze to exist over parts of the Cape Peninsula

to the East of Cape Town. This “Brown Haze” is characterised by a white to brown mist which is trapped at a low level, mostly over the Cape Flats. A study, which focussed mainly on airborne particulate matter, has been performed by the Energy Research Institute of the University of Cape Town, comprising a source apportionment [2,9]. The pilot study [9] had indicated that particulate matter typically has higher light extinction factors than gaseous pollution, and thus the main study [2] focussed most of the effort on particulates. The conclusions drawn in the study were that the most important contributors to the visibility degradation were vehicles: 65% of the light extinction was directly attributed to traffic related causes. Diesel vehicle particulate emissions were found to be the major offender with 48%. Therefore, 17% of the visibility degradation was directly associated with petrol fuelled vehicles.

Visibility impairment by way of light extinction can be divided into four distinct categories:

• light scattering by particles

• light absorption by particles

• light scattering by gases

• light absorption by gases

The role of particles is indicated to be more important in direct visibility degradation than gases, however,

NO2 is known to be a gas which does produce significant visibility degradation and is a product of

photochemical smog [9]. Elevated concentrations of NO2 in ambient air are known to result in a brown

discolouration.

Gaseous pollutant data are routinely monitored in and around Cape Town by the Scientific Services and the Air Pollution Control divisions of the Cape Metropolitan Council. Studying examples of data as given below in Figure 2-3 and Figure 2-4, the conclusion may be drawn that vehicle related photochemical activity is prevalent. As stated above, the combination of oxides of nitrogen and hydrocarbons in the presence of ultra violet light will, in certain circumstances, lead to the formation of photochemical smog. One of the products of this reaction is ozone. It is seen below in Figure 2-3 that on the three consecutive

days (Wednesday 19th to Friday 21st June 1996) there are distinct trends of high NO and NMHC peaks at

around 8am. There are also lower, but wider spread peaks in late afternoon or early evening. These trends are not consistent with industry or household activity but are consistent with commuter traffic patterns which indicate that these pollutants are probably dominated by passenger vehicles. Furthermore,

on two of these days, the 19th and 21st, there is a distinct NO2 peak after the morning’s elevated NO levels.

This is occurring as indicated by reaction 1 in Figure 2-2. This NO2 may then, in the presence of ultra

violet light, decompose and form NO and ozone as indicated by reactions 2 and 3. Furthermore, the presence of the NMHC provides the other reagent to promote the formation of photochemical smog. A

typical photochemical smog episode is seen below in Figure 2-4, where on Sunday the 19th of May ozone

levels are seen to be over the World Heath Organisation guideline. Again the morning and

afternoon/evening peak in NO is seen, followed by NO2 formation and the production of ozone. Therefore,

:

Town environs. It must be stressed, that the photochemical smog chain reaction will not always occur

given, NO, NMHC and ultra violet light. In fact, there was no related ozone formation on the 21st June

1996 due to the extremely high NO levels, in which case extreme NOx inhibition is occurring. However, it

is apparent that there are occasions when the conditions favour this occurrence, and it appears that they are dominated by passenger vehicle emissions.

Figure 2-3 Atmospheric pollutant concentrations for City Hall monitoring site, Cape Town for three consecutive days in June 1996 (graphs courtesy of Scientific Services Division of CMC).

Figure 2-4. Atmospheric pollutant concentrations for Goodwood monitoring site, Cape Town for two consecutive days in 1996 (graph courtesy of Scientific Services Division of CMC).

2.2.2.

Holistic Approach to Air Quality Impact

A holistic approach is required when considering the environmental, and specifically the air quality, impact of vehicle exhaust emissions. Firstly, the exhaust emissions are not the only vehicle related emissions that can influence air quality. Evaporative emissions and road dust [2, 9] are also known to play a role in ambient air quality. Secondly, it is vital to consider exhaust emissions in a rationalised way. In other words, emissions should be quantified in units that take into account influences on engine efficiency and other operating parameters. Exhaust emissions are usually measured in units of concentration of exhaust gas (% by volume or parts per million, ppm) but engines, especially SI engines, have highly variable mass throughput rates. Thus, using units of concentration may be misleading as far as actual air quality impact is concerned, and these results should be rationalised before reliable conclusions can be drawn.

Units of grams of pollutant per kilometre travelled (g/km) or grams of pollutant per kilowatt-hour of engine operation (g/kW.hr) are examples of appropriate rationalised units. By way of example, reducing the

engine compression ratio may drastically reduce the NOx concentrations in the exhaust stream, but there

is an associated reduction in engine efficiency. In order for this engine to do a similar amount of work, it would have to consume more fuel and air. Thus the actual mass of pollutant emitted may actually be

increased relative to the engine with the higher compression ratio and higher NOx exhaust concentrations.

Furthermore, the reduced compression ratio may have impacts on other pollutants emitted, which need to be considered.

%%

processes requiring excessive energy are required, then the displaced emissions of this energy need to be considered. The overall cycle efficiency, or well-to-wheels efficiency, needs consideration too [10]. Another point that bears consideration is that there are many other ways of reducing the environmental impact of vehicles and transport [11] and the potential for reduced impact may be much larger than that achievable only from vehicle technology and fuel formulation. Reducing the number of vehicle kilometres travelled in a given zone will immediately reduce the release of pollutants. This can be achieved by encouraging the average vehicle occupancy to be raised by increasing the use of public transport services such as mini-bus taxis and commuter buses as well as car pooling which may be promoted by increasing parking fees and providing special allowances to high occupancy vehicles. Encouraging the use commuter rail as a convenient alternative could have a significant impact also. Improvements to traffic flow rates may at first seem to be desirable in that it would reduce the trip specific emissions. However, care must be taken as it has been reported that in areas where this has been achieved, the increased traffic efficiency has encouraged more vehicles to travel into the zone and thus actually increasing the release of pollutants [12].

2.3.

Exhaust Gas Composition

The main constituents of an SI engine’s exhaust gas are Nitrogen (N2), Carbon Dioxide (CO2), water (H2O),

Carbon Monoxide (CO), Oxygen (O2), Unburned Hydrocarbons (HC’s) and Oxides of Nitrogen (NOx). Aldehydes

and other partial oxidation compounds are released in much smaller concentrations. Sulphur dioxide is emitted from SI engines in extremely low concentrations due to the generally low concentration of sulphur in the parent

fuel. Diesel engines do however release some SO2 due to the often higher levels of sulphur in the diesel fuel.

Particulates are also not emitted by conventional SI engines in significant quantities. The compounds of interest

are CO2, a greenhouse gas, CO a toxin and NOx and HC’s which are precursors to photochemical smog. The

gases covered internationally by legislation for vehicles propelled by SI engines are NOx, Hydrocarbons and CO,

and this group of gases are referred to as the “regulated emissions”. Particulates would be included in the definition of regulated emissions for diesel propelled vehicles.

The complete combustion of hydrocarbon based fuels produces only CO2 and water. CO is produced if the

combustion is not allowed to go to completion or if there is insufficient oxygen for complete combustion (a rich mixture). Hydrocarbons are present due either to combustion being incomplete or if some of the fuel or mixture does not take part in combustion, hence the term unburned hydrocarbons is often used. Excessive hydrocarbon emission is therefore often due to a rich mixture. Oxides of nitrogen are formed as a natural by-product due to the high temperatures and pressures associated with combustion in Internal Combustion Engines (ICE),

promoting the dissociation of N2 and the resulting formation of NO and other oxides of nitrogen.

The exhaust constituents of interest for this research are the regulated emissions. Exhaust aldehydes were measured and speciated alongside in a parallel project [13,14], but will not be dealt with here.

2.3.1.

Emissions Formation Mechanisms

In order to understand the effect that engine or fuel properties may have on the emissions produced, it is necessary to understand the formation process and origin of the emissions. Thus the formation of each of the gases is discussed briefly below, along with the factors that may affect them.

Carbon Monoxide (CO)

An engine which is operating with an equivalence ratio richer than stoichiometric, will produce CO emissions due to the fact that there is insufficient oxygen available to convert all the carbon in the fuel to

CO2. The chemistry of the C-O-H system in SI engines is generally fast enough that local equilibrium may

be assumed. It may therefore be assumed that lean running engines will produce no CO emission and the amount of CO produced by a rich running engine could easily be determined from a solution of the combustion equation or an equilibrium subroutine. However, once gas temperatures drop during expansion the C-O-H chemistry does become rate limited and therefore equilibrium can not be assumed [15]. Thus wall quenching and the lower temperatures occurring late in the expansion stroke will lead to the appearance of some CO in mixtures running lean of stoichiometric. The highly non-linear nature of CO formation as a function of equivalence ratio around the stoichiometric point, as evident from Figure 2-5 below, leads to a further significant source of CO emission. In a multi-cylinder engine, running with an overall equivalence ratio near stoichiometric, with some cylinders run lean and some cylinders run rich, the lean cylinders produce much less CO than the rich cylinders. The actual CO emissions would be higher than that predicted given the overall operating equivalence ratio. Therefore a multi-cylinder engine running near to stoichiometric but having even slightly uneven fuel distribution will produce disproportionately high CO emission. It thus apparent that the operating equivalence ratio is by far the most significant factor affecting CO formation and uneven fuel distribution will have a disproportionate effect on CO emissions.

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.7 0.8 0.9 1 1.1 1.2 Equivalence Ratio, φφφφ [-] M ol e Fr ac tio n [-] CO2 H2O O2 CO H2

Figure 2-5 Mole fractions of combustion products as a function of equivalence ratio from basic stoichiometry and equilibrium calculations.

%/

Hydrocarbons (HC’s)

The specific hydrocarbons found in the exhaust gases comprise of both compounds found in the parent fuel and compounds not present in the fuel [15]. The latter may be derived from the fuel, the structure having been altered within the cylinder by chemical reactions which did not go to completion, or a small amount resulting from the hydrocarbon based lubricant. An engine running rich of stoichiometric would have unburned fuel present in the exhaust stream due to insufficient oxygen being available during combustion for complete combustion to take place. However, even stoichiometric and lean running engines have some hydrocarbons present in the exhaust. There are three important mechanisms which are believed to be responsible for the presence of these hydrocarbons in the exhaust stream:

1. misfire

2. storage and release of fuel by deposits and oil layers or flame quenching near the walls 3. flame quenching within crevice volumes.

Any form of misfire, whether total lack of combustion or partial combustion is usually associated with engines in a poor condition or operating under unusual conditions (for instance cold operation). Thus this mechanism for the presence of hydrocarbons is not of major concern in this research.

The remaining two mechanisms are closely linked and will be discussed together. In both cases the progressive flame is inhibited from reaching, and therefore burning, an amount of the inlet charge, which is then exhausted during blow-down. During compression, as the pressure increases, mixture is forced into the region between the top piston ring, piston and cylinder wall, called the top land crevice volume. Other crevice volumes are also subject to this filling procedure and they consist of the ‘caves’ formed by the rough nature of engine deposits and any other small gaps such as the protruding thread of the spark plug. The dimensions of these crevice volumes are smaller than the flame quenching distance and this prevents the passage of the flame into this region. Increased pressures during compression and combustion also promote the solubility of the fuel into the oil layer. During expansion, and when the exhaust valve opens and pressures begin to fall, the mixture in the crevice volumes and deposits expands and re-enters the cylinder while the fuel absorbed in the oil layer is also released. Some of this fuel may undergo combustion or partial combustion, but as these volumes of mixture are in close proximity to the cylinder surfaces and it is late in the cycle, their temperatures are not high enough to produce rapid reaction rates or combustion.

Some researchers [16] doubt the role of the oil layer absorption playing a role and argue that the flame is quenched as it approaches near to the relatively cold cylinder wall, therefore precluding a portion of the mixture from combustion. The exact mechanism taking place is not of major concern as both mechanisms are influenced by the same variables. The amount of fuel mixture involved in either mechanism is dependent on the same factors, they are both proportional to the surface area exposed and dependent on the cylinder pressure time history. Higher the cylinder pressures increase the density of the gas in the quench layer and thus there is a greater tendency for fuel absorption into the oil layer. However, if oil layer absorption is a valid mechanism then the specific fuel/oil solubility will influence the process.

Furthermore, the relative amount of fuel present in the gas which is forced into the crevices or is present in the quench layer, will be dependent on the equivalence ratio of the gas and in-cylinder mixing. Therefore the overall operational equivalence ratio, as well as mixing can influence this. It is possible that under certain conditions, possibly poor mixing or liquid fuel impingement on the cylinder surfaces, the equivalence ratio of the gas in these regions may differ from the bulk gas equivalence ratio.

It is important to note, too, that there is a considerable amount of burn-up occurring in the exhaust port. The bulk exhaust temperatures may still be high enough to promote some oxidation and the turbulence caused by the exhaust valve causes significant gas mixing. Thus the mixture that was precluded from combustion by contact with the relatively cold surfaces within the chamber becomes mixed with the hotter bulk gas, and some further combustion may take place.

Therefore the factors that effect the emissions of hydrocarbons include the equivalence ratio, crevice volumes, and flame quench area, cylinder pressure time history, the amount of post combustion burn-up and, to a lesser extent, fuel/oil solubility.

Oxides of Nitrogen (NO

x

)

Oxides of Nitrogen are not a direct product of the combustion reaction, but are formed due to the high

temperatures and pressures within the cylinder promoting the dissociation of N2 to monatomic nitrogen.

This nitrogen can then be oxidised by the oxygen present in the cylinder thus forming NO and, to a lesser degree higher oxides. Three distinct NO formation mechanisms can be identified, the thermal, prompt and nitrous oxide mechanisms (15, 17, 18).

For the thermal mechanism, the chemistry associated with the reactions involved in the formation of NO, in contrast with the C-O-H system, is not fast enough to be assumed to be in equilibrium, but is rate limited. In 1946, Zeldovich identified the most important reactions relevant to the thermal mechanism and thus this mechanism is often referred to as the Zeldovich mechanism (15, 17, 18). The rate of one of the important chemical reactions, that of the combining of monatomic nitrogen with oxygen forming NO and monatomic oxygen is highly temperature dependant. This implies that the actual concentrations lag behind the concentrations that would be predicted by equilibrium concentrations. This is demonstrated below in

Figure 2-6. As can be seen from Figure 2-6 (c) the actual concentrations tend to equilibrate, if equilibrium

concentrations are higher than actual concentrations then nitric oxides are being formed. If actual concentrations are higher, then the nitric oxides decompose. It is also important to realise that the chemical reaction rates become low enough at temperatures below 2000 K to assume that decomposition of nitric oxides is negligible and then the concentrations are assumed to be frozen. The net result is that the level of nitric oxides in the exhaust stream is higher than those predicted by equilibrium calculations. The Zeldovich mechanism for NO formation is presented in detail in Section 8.3.5 below.

%4

Figure 2-6 (a) Measured pressure P and calculated mass fraction burned, x as a function of crank angle. (b) Calculated temperature of burned gas Tb and unburned gas Tu as a function of crank angle for two

elements that burn at different times. (c) NO mass fractions as function of crank angle for two elements that burn at different times (From Ferguson [17])

It can also be seen from Figure 2-6 (b) that the temperature histories of each element to burn are different.

This occurs because each element to burn does so at a different time and is therefore at a different temperature at the time it burns (due to the continual compression ahead of the flame propagation). The element then burns to its adiabatic flame temperature and then tracks pressure as it is more or less isentropically compressed and expanded. As a result of this the first element to burn attains the highest temperatures as it is compressed after burning. Conversely, the last element to burn has the lowest temperature time history as it only expanded after burning. This also leads to different NO formation paths for elements burning at different times, as indicated in Figure 2-6 (c). It is apparent therefore that the thermal mechanism, due to low initial reaction rates, only forms NO in the hot burned gas.

NO is found to be present in the flame, and therefore can not have been formed through the thermal mechanism. The prompt mechanism is identified as being relevant here. The prompt mechanism relies

on reactions in which N2 and hydrocarbons react, liberating monatomic nitrogen which can then oxidise.

The prompt mechanism is important if there is fuel bound nitrogen, and the resultant concentrations are much less than that associated with the thermal mechanism unless the burned gas temperatures are so low that the thermal mechanism is itself not significant. The nitrous oxide mechanism is significant for lean

The thermal mechanism is most significant for spark ignition engines, while the prompt mechanism becomes important for compression ignition engines.

Engine operating conditions can influence the NOx concentrations in the exhaust stream by way of

affecting the temperature time histories of the cylinder charge. Equivalence ratio can influence in-cylinder gas temperatures as well as oxygen availability. Lean mixtures have more air for a given fuel mass induced, while rich mixtures have excess fuel. In both cases there is more mass to absorb heat per fuel energy content, thus resulting in lower charge temperatures than for stoichiometric conditions. In actual engines, however, maximum charge temperatures actually occur at equivalence ratios slightly rich of stoichiometric. This is due to the fact that imperfect charge mixing occurs, and it is difficult to burn, and

release the energy of all of the fuel induced. Maximum NOx emissions are actually usually found slightly

lean of stoichiometric where temperatures are still high but there is now sufficient oxygen available for the reactions to progress effectively.

Ignition timing can significantly influence charge temperature histories. This is due to the direct effect that ignition timing has on cylinder pressure history. Ignition advanced from Maximum Brake Torque (MBT) will tend to produce higher maximum pressures because the energy release to the charge will raise the cylinder pressure, which is then compressed, further raising the pressure. Opposed to this, ignition retarded from MBT will tend to have lower maximum pressures because cylinder expansion will begin earlier in the heat release process. The corresponding charge temperatures will similarly be influenced,

and in this way alter the formation of the rate limited reactions involved in the formation of NOx.

Compression ratio too influences charge temperature histories. Higher compression ratios increase the cylinder pressures and temperatures throughout the cycle and would thus be expected to produce

increased NOx exhaust concentrations. The rate of combustion too can influence temperature time

histories where increased burn speed would be expected to raise maximum temperatures.

2.4.

Exhaust Emissions Measurement

Before continuing with a survey of studies previously undertaken, it is important to outline the methods used to measure exhaust composition. There are a number of ways to accomplish this, with varying degrees of sophistication and therefore effort required. These different methods also provide different qualities of information. These are outlined below. All of these methods rely on similar analysis equipment for the determination of the concentration of pollutants, but it is the method of gathering the exhaust sample for analysis that differs. Detail of the analysis equipment is therefore not included here.

2.4.1.

Steady State Engine Operation

This is the simplest method of capturing an exhaust sample for analysis. An engine is operated on a bench dynamometer at a steady speed and load. A continuous sample of exhaust gas is drawn from the exhaust and passed to the analysis equipment. Once the engine and analysis equipment has stabilised,